本論文提出一個心電訊號處理與心肌梗塞疾病辨識系統設計。整體設計的目的是實現具精準度、即時性、簡易又快速之心血管疾病自動判別系統,提供心肌梗塞症狀之預警機制。在本研究中,針對心電圖中的肌電雜訊進行處理,採用Butterworth高通濾波器來移除。心肌梗塞發生初期,心電圖中ST段會發生異常偏移,故我們針對此特徵來開發系統。疾病辨識系統中,本研究提出將S波至T波平均取六點,選用第二到第四點做為ST段偏移判別的資料。此方法能降低ST段檢測失誤,以及增加辨識準確度。除此之外,以ST段這三點數據搭配型態學分析,將ST段分成圓頂型、凹口型、直線型,分類的結果將有助醫療人員觀察疾病症狀。 本研究中使用MIT-BIH與歐洲ST-T資料庫,首先在MATLAB中開發與驗證辨識系統,再轉移到FPGA中設計證實可行性,最後結合LabVIEW顯示診斷結果。經過消除肌電雜訊機制後,本系統疾病辨識敏感度為96.6%、特異度為93.1%。
This thesis presents a systematic design of ECG signal processing system for detections of myocardial infarctions. The goal is to implement a automatic identification system for cardiovascular disease with high accuracy. To provide early warning. We implemented a butterworth high-pass filter to effectively remove EMG interferences. ST segment of ECG will be pathologic shift in early myocardial infarction. This thesis presents average take six point from S wave to T wave, and choice of second to fourth point as ST segment deviation diagnostic data. This method can reduce ST segment detection errors, and improve identification accuracy. Moreover, we implemented a morphological analysis method with these three data of ST segment. The ST segment is divided into convex, concave and straight. The results of the classification will help medical personnel to observe symptoms of disease. The test data are from the MIT-BIH arrhythmia and European ST-T database. System validation is carried out in MATLAB, and DSP algorithms are implemented in the FPGA. Diagnostic results can be illustrated on LabVIEW. Experimental results show that the sensitivity and specificity of ST segment classification after EMG elimination is 96.6% and 93.1%, respectively.